DI
SC
US
SI
ON
P
AP
ER
S
ER
IE
S
Forschungsinstitut zur Zukunft der ArbeitInstitute for the Study of Labor
Bride Price and Fertility Decisions:Evidence from Rural Senegal
IZA DP No. 7770
November 2013
Linguère Mously MbayeNatascha Wagner
Bride Price and Fertility Decisions:
Evidence from Rural Senegal
Linguère Mously Mbaye IZA
Natascha Wagner ISS, Erasmus University Rotterdam
Discussion Paper No. 7770 November 2013
IZA
P.O. Box 7240 53072 Bonn
Germany
Phone: +49-228-3894-0 Fax: +49-228-3894-180
E-mail: [email protected]
Any opinions expressed here are those of the author(s) and not those of IZA. Research published in this series may include views on policy, but the institute itself takes no institutional policy positions. The IZA research network is committed to the IZA Guiding Principles of Research Integrity. The Institute for the Study of Labor (IZA) in Bonn is a local and virtual international research center and a place of communication between science, politics and business. IZA is an independent nonprofit organization supported by Deutsche Post Foundation. The center is associated with the University of Bonn and offers a stimulating research environment through its international network, workshops and conferences, data service, project support, research visits and doctoral program. IZA engages in (i) original and internationally competitive research in all fields of labor economics, (ii) development of policy concepts, and (iii) dissemination of research results and concepts to the interested public. IZA Discussion Papers often represent preliminary work and are circulated to encourage discussion. Citation of such a paper should account for its provisional character. A revised version may be available directly from the author.
IZA Discussion Paper No. 7770 November 2013
ABSTRACT
Bride Price and Fertility Decisions: Evidence from Rural Senegal*
This paper is the first to provide evidence about the relationship between bride price payments and fertility decisions in the African context. Remarkably, the results show that bride price payments reduce fertility pressure, with a woman reducing her number of children by 0.5 at the mean bride price. The results are robust to different tests that we conduct to address the potential endogeneity between bride price payments and fertility decisions. As possible transmission channels, we find that poor women and men with low levels of education are the most negatively affected by the tradition of bride price payments. Furthermore, a lower bride price payment increases fertility pressure in polygamous households and for arranged marriages, while the bride price payment has no effect on the couple’s decisions concerning fertility in monogamous households and for love marriages. Consequently, given that bride price payments have less power over (economically) independent women, empowerment will give leeway to girls in traditional societies, even if the bride price system is not overturned. JEL Classification: O12, Z13 Keywords: bride price, empowerment, fertility, marriage payments, Senegal Corresponding author: Linguère Mously Mbaye IZA P.O. Box 7240 53072 Bonn Germany E-mail: [email protected]
* We are grateful to Alpaslan Akay for valuable comments and suggestions on earlier versions of this paper. We also thank Kosali Simon, seminar and conference participants at the Midwestern Econometrics Group Meeting 2013, Indiana, USA; the 8th IZA/World Bank Conference on Employment and Development, Bonn, Germany; the 18th Annual Conference on Econometric Analysis and Policy Challenges in Africa, Accra, Ghana; the 88th Annual Conference of the Western Economic Association, Seattle, USA; the IZA Brown Bag Seminar, Bonn, Germany; and the 27th Annual Conference of the European Society for Population Economics, Aarhus, Denmark. We also acknowledge the participation of Jean-Louis Arcand, Marie-Charlotte Buisson and Samba Mbaye in the survey design and data collection. The usual disclaimer applies.
1 Introduction
The bride price payment is a key element of the marriage contract in many sub-
Saharan African countries, and particularly in Senegal. Contrary to a dowry pay-
ment, where the bride and her family give the marriage payment to the groom
and his family, a bride price payment is defined as the money or wealth transfer
given by or on behalf of the groom to the bride and her family upon the mar-
riage of the couple.1 Such a bride price system raises several concerns related to
the economics of the marriage market, gender empowerment and intra-household
bargaining power. We consider a woman’s role in marriage as represented by the
fertility decisions of the couple. Previous research for the case of Indonesia has
shown that women who perceive more household assets as being their own have
a greater say in fertility decisions (Beegle et al. 2001). Thus, fertility choices, as
measured by the number of children at a given age, indicate female bargaining
power and a woman’s relative position within the household.
This paper provides an important contribution to the existing literature, as most
studies about marriage payments have been carried out in South Asia. However,
the African context of a bride price being paid is substantially different and the
dynamics that apply in many parts of Asia cannot be extrapolated to the African
continent. To the best of our knowledge, our study is the first large-scale analysis
of bride price dynamics in Africa and the first that is interested in the relationship
between bride price payments and fertility decisions in general, and in rural Africa
in particular.
The ancient practice of bride price payments is performed in many sub-Saharan
African countries.2 In the specific case of Senegal, the value of the bride price
1In Senegal, the bride price is given to the bride and her family. In this context, it is impossibleto disentangle the share of the marriage payment that goes to the bride herself from the sharekept by her family. However, it is important to bear in mind that in African societies, andparticularly in traditional societies, the group is more important and valued than the individual.Consequently, the specific share received by the bride is of minor importance, given that shemakes part of her family of origin. What is of greater importance is the transaction fromone family to another. Whenever the family of the bride receives a transfer, the bride will beconsidered as the one who received the payment and was valued by her in-laws, regardless of theshare of the transfer over which she has direct control.
2The following sub-Saharan African countries practice bride price payments: Ghana, Kenya,Nigeria, Rwanda, Senegal, South Africa, Tanzania and Uganda. The list is non-exclusive. Theaforementioned countries participated in an international conference about the system of bride
is determined in negotiations between the two families. The bargaining process
preceding a marriage might only involve the couple or even take into considera-
tion advice given by the extended family. Before the actual marriage celebration,
several consultative visits and mutual invitations take place between the future
spouses and/or close family members. To conclude this period of negotiations, the
bride’s parents have to choose a date for the marriage ceremony and announce
the desired value of the bride price, which can comprise cash and material goods
such as furniture and jewels. Moreover, most of the Senegalese are Muslims and,
according to the Islamic law, a minimum amount is necessary to validate the
marriage. However, this amount is suffi ciently low to only be of symbolic value
(around FCFA 5,000 or EUR 8). It is common practice that people give a higher
bride price than what is required by religion. Moreover, in sub-Saharan African
countries, bride price payments result from tribal traditions and are practiced re-
gardless of religious affi liation, with both Muslims and Christians adhering to these
traditions. As a bride price exceeding the amount set by Islamic law is not fixed,
its precise amount depends on the appreciation of the bride by both her family
and in-laws, the bride’s socio-economic conditions, her relative bargaining power,
the prestige of both families as well as customary law. In the extreme case, the
bride price payment asked for by the girl’s family might be prohibitively high to
prevent the union with an unpopular man. In a different context, the girl’s family
might abstain from any claim for altruistic reasons. Given that the actual nego-
tiation is a dynamic process that depends upon the socio-economic conditions of
the two families, the value of the realized bride price payment varies considerably
across couples.3For the purpose of this study, we employ a unique panel dataset of
2,241 observations, which allows us to quantify the effect of the bride price on the
number of children that a woman gives birth to. We find a negative relationship
price payments in Uganda in 2004.3A historical overview of marriage laws in early African societies reveals that marriage pay-
ments have not always been perceived as a "bride price". According to Kanji and Camara(2000), the term "prenuptial gift" is more appropriate than bride price. Traditionally, marriagepayments were not viewed as a price or compensation. Such transfers were of symbolic nature,representing the engagement of the groom towards his bride and her family. Nonetheless, thesenuptial transfers were interpreted as a bride price as time evolved, with the concept furthershaped by traditional, religious and political influences. The aim of our study is not to trigger ananthropological debate about the historical perception of the bride price. However, we considercurrent dynamics and their impact on women’s lives in rural West-Africa.
2
between bride price payments and the number of children, namely that women
who receive a higher bride price have fewer children. Indeed, at the mean bride
price, a woman reduces her number of children by 0.5. Thus, the more a woman is
perceived as precious investment by her in-laws, the lower the pressure upon her
for reproductive success. The results highlight the importance of the bride price
in establishing a socially accepted and enforceable contract. The higher the man’s
investment in his wife, the higher her value and appreciation and consequently the
fewer children she has to give birth to. This suggests that the bride price system
influences the bargaining power and independence of women in a pervasive man-
ner. We address the potential endogeneity issues related to the omitted variable
bias or the measurement error of the bride price through a series of robustness
tests. We demonstrate that our findings are not driven by these sources of bias,
with our results even robust if we control for income, wealth and the economic
conditions at the time of the marriage, or if we use alternative specification. We
also carry out a very detailed analysis by exploring various transmission channels
along which the marriage payments influence the fertility decisions.
The remainder of the paper is structured as follows. Section 2 reviews the exist-
ing literature. In Section 3, we present the study context and the data.Section 4
introduces the empirical strategy and the main results. Endogeneity issues and
robustness checks are discussed in Section 5. Heterogeneous effects and various
transmission channels are presented in Section 6. Section 7 concludes and presents
the implications of our results.
2 Background on marriage payments
Family and marriage dynamics were first put on the agenda of economic research
by Becker (1981), whose model of marriage rests upon the equilibrium of a match-
ing exercise, fitting men and women together as couples. This assortative mat-
ing model can explain the existence of dowries and bride price payments, with
the latter particularly arising in polygamous societies where women are relatively
scarce. These payments allow differentiating between single and divorced women,
for example, with the former being more highly valued on the marriage market
(Goldschmidt, 1974; Papps, 1980). Thus, Becker views the bride price payment
3
as an investment. According to his analysis, bride price payments are forfeited if
husbands divorce their wife without cause. This explains why husbands have the
incentive to maintain the social contract in which they have engaged. In turn,
wives have to repay their ex-husbands if they leave them without any reason.
Marriage payments are a common concept in almost all societies. Despite hardly
ever being practiced in the Western world, they remain prominent in many de-
veloping countries (Andersson, 2007). To date, most of the studies concerning
marriage payments have been carried out in South Asia. In this region, marriage
payments are mostly found in the form of dowries, meaning that the marriage
payment is given by the bride’s parents to the groom. Several studies within the
existing literature have analyzed the relationship between dowry payments and
various other factors such as the surplus of women in the marriage market (Rao,
1993), the cast system (Anderson, 2003), domestic violence (Srinivasan and Bedi,
2007), female human capital endowment (Sharma and Frijters, 2009) and family
planning (Peters, 2011). Botticini and Siow (2003) argue that dowries and bequests
are used in patrilocal societies to avoid free-riding behavior between siblings, as
well as effi ciently transferring wealth from parents to children.
Most African countries base the marriage contract upon bride price payments
rather than dowries.4 A plethora of reasons for this difference in marriage sys-
tems has been discussed (Botticini and Siow, 2003). First, Africa is characterized
by collective rights that prevent a dowry system from evolving, as the possibility
for parents to transfer individual wealth to their children are limited. Second,
Africa has a non-plough agricultural system where female labor is more impor-
tant than in the plough agriculture that exists in Asia (Boserup, 1970). African
agriculture rather destroys the land that is cultivated and moves on to new places
for cultivation, thus limiting the possibilities to introduce dowries and bequests.
Third, dowries are found in monogamous societies, whereas African societies have
a polygamous structure (Bergstrom, 1994; Tertilt, 2005).
The economics of bride price systems has received increasing attention in recent
years. For instance, it has been shown that higher bride price payments do not
4There are also cases in Asia where bride price payments are practiced, such as in Thailand(Cherlin and Chamratrithirong, 1988) or China, where it is possible to find the two systems ofmarriage payments coexisting (Anderson, 2007; Brown, 2009).
4
influence the behavior of husbands towards their wives, whereas higher dowries do
(Zhang and Chan, 1999). In a study using Ugandan data, Bishai and Grossbard
(2010) interpret the bride price as the women’s sexual fidelity to men, because
this marriage payment is associated with a lower rate of extra-marital sexual re-
lationships for women, yet not for men. In the specific case of Senegal, Gaspart
and Platteau (2010, 2007) show that the value of the bride price is determined by
a woman’s socio-economic characteristics, as well as strategic behavior by herself
and her parents. The bride and her altruistic parents can prefer a (relatively) low
bride price in the case of a love marriage in order to reduce the risk of ill-treatment
and the marriage’s failure.
3 Survey context and data
3.1 The survey
Data are provided by a household survey carried out in eight regions of Senegal.5
The survey was part of the program evaluation of a rural electrification initiative
by the UNDP known as a multifunctional platform. The dataset includes two
waves: the baseline survey was carried out between May and July 2009 and the
second survey between April and June 2011. Villages were randomly selected based
on the criterion of not having access to energy. Within the villages, households
were also selected at random from the list of resident households supplied by the
head of the village. The sample is thus representative of rural Senegal in eight of
14 regions in which subsistence agriculture is the most prevalent form of income
generation.
For the purpose of our analysis, we restrict the sample to individuals who were
married at the moment of the survey and responded to the gender questionnaire.
While the majority of these couples are the same across waves, not all couples
were reinterviewed in 2011 as they might have been replaced by another couple
living in the same household. Therefore, we also have new couples that enter
the sample in 2011. Consequently, we establish an unbalanced panel of married
5The regions are Diourbel, Fatick, Kaolack, Kedougou, Kolda, Louga, Tambacounda andThies.
5
couples, which allows us to compare the dynamics of marriage payments across
regions. We noticed among the same couples present in the two waves that the
amount of the bride price recalled by the same woman differed from one survey
round to the next. We use this variation to later address the potential recall bias.
3.2 Descriptive statistics
We present summary statistics in Table 1. The vast majority of the households
(95%) are Muslims and consider themselves as members of the Wolof (43%), Pular
(24%) or Serere (18%) ethnic group. Moreover, almost one-fifth of the households
are casted and 42% of the couples report living in a polygamous union. In the
case of polygamy, first wives represent 26% of our sample.6We proxy for wealth
with landownership, which represents the number of arable plots owned by the
household. On average, households own 2.51 plots, indicating that they are in
agriculture and derive their living from farming. In order to capture the economic
conditions at the time of the marriage, we build five cohorts derived from the
difference between the year of the survey and the year of marriage. For instance,
the year of marriage cohort [0, 5[ is an indicator variable equal to one if the
marriage was celebrated less than five years before the survey, while the year of
marriage cohort [5, 10[ codes for women who were married between five and ten
years before the survey. We construct the year of marriage cohorts [10, 15[, [15,
20[ and [25,.] by following the same logic. 42% of the marriages were celebrated
at least 25 years before the survey.
At the level of the individual couple, we derive most information from the woman.
Women have on average 5.20 children, including both those who are alive and
those who have died. The average woman is almost 37 years old, which makes
her 14 years younger than her husband. This gender age gap is not surprising in
rural Senegal and gives an indication of the timing of marriages and the underlying
decision process: Men first need to acquire the necessary assets to set up a family,
of which the bride price paid at the wedding constitutes a non-negligible part.
6Concerning the ranking of the wives, being the first wife is coded as a dummy equal to oneif the woman is the first wife and 0 otherwise. The same logic applies for the indicator variablessecond wife and third or fourth wife. The excluded category is constituted by monogamouslymarried women.
6
While the average bride price in the sample is FCFA 240,917 (367 Euros), there
is considerable variation and the mean is driven by some large amounts, as the
median of FCFA 75,000 (114 Euros) shows. Women marry young, as the average
age at first marriage indicates: they are below 18 years on average. Moreover,
divorce is rare (8%). Women’s position is weak, with only 16% knowing how to read
and write, whereas 44% of their husbands are alphabetized. Nonetheless, almost
47% of the women in the sample are economically active beyond household chores
and family farm work, while 91% of men pursue additional economic activities.
Women’s subordinate position is also demonstrated with respect to their living
arrangements. Women typically live close to their in-laws, with the couples living
on average only 2.55 km away from the husband’s parents. By contrast, this
distance is estimated at 20 km for the woman’s parents. This last pair of descriptive
statistics represents virilocality in the rural communities where wives have to take
care of their parents in-law and the domestic activities in the in-laws home. Last
but not least, these descriptive statistics also show that the majority of couples
come from the same region, as indicated by the relatively short distances to the
parents in general.
Overall, summary statistics leave no doubt that the sample at hand represents
poor rural villages in Senegal, within which women are in a considerably weaker
position than men in terms of education and (economic) empowerment.
4 Econometric framework and main results
4.1 Empirical strategy
The linear regression framework is employed to analyze the link between bride
price payments and fertility decisions. We consider the following empirical model:
yihrt = β0 + β1 log(BP )ihrt + β2Wihrt + β3Hihrt + β4HHhrt + µt + λr + εihrt, (1)
where the subscript i denotes wife, h household, t year and r region. The de-
pendent variable yihrt is the number of children a woman has given birth to at
7
the moment of the survey. The variable of interest is the logarithm of the bride
price log(BP )ihrt. We also control for the socio-demographic characteristics of the
women Wihrt such as age and its square, the age of the women at her current
marriage and its square, as well as a binary variable equal to one if the women
is literate. The variable literacy is used to proxy for the education level. Sim-
ilarly, we control for the husband’s characteristics Hihrt such as his age and its
square, as well as a binary variable equal to one if he is literate. In addition,
we take into account household characteristics HHhr by controlling for ethnicity,
religion and caste as well as the type of marriage practiced. The marriage form is
captured by an indicator variable coding for polygamous unions. Controlling for
wife-, husband- and household-specific characteristics allows us to address issues of
omitted variable bias at the level of the individual couple. In addition, we include
year of survey dummies µt in our basic regression model to capture time-specific
effects.
Regional heterogeneity is taken into account by including regional fixed effects
λr in equation (1). By doing so, we can rule out sources of bias stemming from
regional characteristics that are common across couples residing in a particular
region. Finally, the unexplained residual is captured by εihrt. Standard errors are
clustered at the regional level to account for intra-cluster correlations.
In the empirical analysis, we will demonstrate the robustness of our results by con-
trolling for other sources of endogeneity. Furthermore, we will control for different
sources of heterogeneity, which will also shed light on transmission channels.
4.2 Main results
The results clearly show that the bride price is negatively related to the number
of children. Whenever the groom gives a higher bride price, he demonstrates his
willingness to treat his wife well and avoid the risk of divorce, which would corre-
spond with a financial failure of his investment. In consequence, a husband who
has paid a higher bride price gives greater independence to his wife by reducing the
pressure on her to produce offspring. In turn, this increases the wife’s bargaining
power in the relationship and the household, as she is more valued as a person.
We start the analysis by comparing plain vanilla OLS results with results account-
8
ing for regional heterogeneity. Table 2 presents the results. Controlling for wife
and husband characteristics jointly, we find a significant and negative relationship
between the bride price and the number of children. In Column 1 of Table 2, the
OLS framework yields a significant and negative coeffi cient associated with the log
bride price payment. Controlling for polygamy, ethnicity, religion and caste in the
OLS specification (Column 2 of Table 2) renders the coeffi cient associated with
bride price insignificant ( p - value of 14.8%). However, the coeffi cient remains
significant at the 5% level when employing regional fixed effects (Column 3 and 4
of Table 2). Nonetheless, it drops in size, indicating that socio-cultural and tradi-
tional factors such as ethnicity, religion and caste are important determinants of
fertility (Column 4 of Table 2). At the mean bride price, the number of children
a woman gives birth to is reduced by 0.5. At the median bride price, a women
has 0.45 children less on average. Put differently, every other women who receives
a bride price between the median and the mean value of our sample gives birth
to one child less. Given that this is an ex post analysis relying on observed fer-
tility outcomes, we control for the age of the husband and wife, as well as their
age at marriage. Unsurprisingly, older women have given birth to more children,
although this relationship is non-linear. The turning point is at the age of 50,
which corresponds to the average age of natural menopause according to medical
research. At the same time, we observe that the younger a bride at her marriage,
the more children she has given birth to at the date of the survey. However, the
relationship between the age at marriage and the fertility decisions is U-shaped,
with 34 years marking the threshold. Similarly, literacy has the expected negative
sign, suggesting that more educated women tend to have fewer children. Nonethe-
less, the significance of the coeffi cient associated with literacy vanishes once we
control for regional fixed effects. This can be explained by the fact that within
the sample at hand, the education level across women is low, resulting in little
variation across women (Columns 3 and 4 of Table 2).
While we do not find any significant differences in fertility patterns across religious
groups or castes, the ethnic group of the Serere tend to have one more child on
average. This result has also been found in the Demographic and Health Surveys
and confirms the representativeness of our dataset for rural Senegal.
Taken together, the impact of the covariates on fertility decisions emphasizes the
9
contextual dimension of both marriage arrangements and ensuing family planning.7
The analysis demonstrates that the individual characteristics of husband and wife
shapes fertility decisions, as does the timing of their marriage. In what follows,
we test the robustness of our results by considering income, wealth and economic
conditions, as well as using some alternative specification. We also explore the
extent to which the couple’s characteristics are linked to bride price payments. By
conducting a differential analysis, we detect transmission channels for the bride
price and improve our understanding of the institutional settings of the marriage
market in rural Senegal.8
5 Dealing with endogeneity
In our econometric approach, we use the detailed information provided by the
household survey and control for the characteristics of the couple and the house-
hold. We also rule out sources of biases that possibly originate from time-invariant
regional characteristics and common trends through the region and survey year
fixed effects, respectively. Despite paying attention to all these elements, we might
still face endogeneity problems due to the omission of relevant variables such as
income and wealth. At the same time, income and wealth themselves are con-
sidered endogenous. However, even if these variables tend be endogenous, by not
controlling for them we might introduce omitted variable bias. Therefore, it is
useful to compare our results with and without income variables. At the very
least, this allows us to ascertain whether our findings are robust to their inclusion.
Moreover, we have information about the landownership status of the household,
which serves as a measure of wealth. Given that this variable is a good proxy for
permanent income and can be considered exogenous due to the lack of active land
markets, we also include it in our regressions. Moreover, we also consider the eco-
nomic conditions at the time of the marriage by including year of marriage cohort
7The results are robust for simple clustering and multi-way clustered standard errors. Thelatter option allows us to account for the small number of clusters (Cameron et al., 2011).
8Given that our dependent variable, number of children, is a count variable, it might beargued that the Poisson model is more suited for the analysis. We also employed the Poissonmodel and derived qualitatively similar results. These findings are not presented for the sake ofbrevity.
10
effects. These marriage cohort dummies control for two sources of bias. First,
they allow us to address the possible omitted variable bias concerning economic
conditions at the time of the marriage. At the same time, they address the second
source of endogeneity introduced by the measurement error in the amount of the
bride price due to the recall nature of the data. As with all analyses that employ
recall data, we face the challenge that human memories are adaptive. Therefore,
it is likely that the precise amount of the bride price is not correctly expressed
by the women interviewed, with variations possibly emerging due to the duration
between the year of the survey and the year of the marriage. One way of capturing
this effect is to control for the year of marriage. Finally, in order to ensure that
the measurement error does not affect the robustness of our results, we use an
alternative specification that considers the log of the mean bride price between
2009 and 2011 reported by each woman who was interviewed twice.
5.1 Controlling for income and wealth
In order to disentangle the income effect from the bride price, we separately include
income generating activities (IGA) undertaken by the wife and husband in our
empirical specification. The variable is binary and equal to one if the wife (the
husband) is economically active. By including IGAs, we control for the labor
market participation of women and men, which can influence both the amount of
the bride price and the number of children. We observe that the wife’s income
generating activities increase her number of off-spring, while such activities by the
husband have no impact on fertility decisions (Column 1 of Table 3)). This result
demonstrates that women provide most of the resources for raising their children
themselves, with paternal contributions being limited (Desai, 1992).
We have not yet accounted for wealth in the analysis, although assets and wealth
clearly influence both bride price payments and fertility decisions. Therefore, we
include landownership as a control variable in the next specifications. This is
considered a suitable wealth indicator as inheritance is patrilineal in rural Senegal
and no active land markets exist.9 In Column 2 of Table 3, we control for the
9Our dataset also contains information about women who bring land into the marriage. How-ever, given that this group of women represents less than 4% of the sample, the variation infemale landownership is too small to be relevant for differences in personal wealth and thus we
11
IGAs of the wife and her husband jointly with landownership, while in Column
3 we consider landownership as the only wealth indicator. Despite controlling for
wealth, socio-cultural factors and regional heterogeneity, the negative relationship
between bride price payments and the number of children a woman gives birth
to remains. The wealth effect is non-negligible and is represented by the positive
impact of landownership on the number of offspring.
5.2 Controlling for the economic conditions at the time of
the marriage
Economic conditions at the time of the marriage are considered underlying factors
that can influence both the amount of the bride price and the ensuing number of
children. In order to deal with this source of omitted variable bias, we include year
of marriage cohort fixed effects in our empirical specification, which allows us to
take into account economic variables such as the level of prices or inflation rates,
as well as positive or negative shocks such as weather events that prevailed around
the wedding period. By employing five year windows, we rule out any impact of
the economic context at the beginning of the marriage on both the bride price
negotiations and the initial fertility decisions.
Moreover, as previously mentioned, the year of marriage cohort fixed effects also
address the measurement error of the bride price payment by capturing the effect
of time on the memories of the women interviewed. Column 4 of Table 3 shows the
results with year of marriage cohort fixed effects. While the coeffi cient associated
with bride price remains significant and negative, its magnitude is slightly lower
compared to Column 3 of Table 2, indicating that the overall economic conditions
around the wedding play a role in ensuing fertility decisions.
5.3 Alternative specification
Although we control for the economic conditions at the time of the marriage, we
want to ensure that the measurement error resulting from the recall of the bride
price does not invalidate our results. We note that the average bride price in
exclude this variable from the empirical analysis.
12
2009 was different from that in 2011, on average by FCFA 128,106.9 (195 Euro),
which is significant at the 1% level (Table 4). The mean bride price between these
years is evaluated on average at FCFA 263,343.3 (401 Euro). Consequently, we
implement an alternative specification that uses the log of the mean bride price
between 2009 and 2011 rather than the actually reported values in every year. In
Columns 1 to 4 of Table 5, we run the same estimations as in Table 2. Region fixed
effects results are similar in terms of significance and sign, and the results remain
unchanged even when we control for landownership and year of marriage cohorts
(Column 5 of Table 5): The relationship between bride price and the number of
children remains significant and negative. The only difference between the main
results and those using the mean bride price is the size of the coeffi cients, whose
magnitude is higher in absolute terms when we use the mean bride price rather
than the initial values. Consequently, we prefer to be conservative in assessing the
magnitudes and thus focus on those in Table 2, which are lower in absolute terms.
Given that these various robustness checks show that our results are not driven
by an omitted variable or a measurement error bias, we proceed further in our
analysis by exploring possible transmission channels that can explain the negative
relationship between the bride price and fertility decisions.
6 Heterogeneous effects and transmission chan-
nels
To gain a more precise picture of the differential impact of bride price payments
with respect to human capital and wealth, we try to carve out potential transmis-
sion channels along which these nuptial payments affect fertility decisions. There-
fore, we introduce various interaction terms between the bride price and human
capital, as well as wealth and income.
13
6.1 Heterogeneity with respect to human capital and in-
come
First, we begin with the literacy of the wife. In Column 1 of Table 6, we compare
the coeffi cient estimates associated with the differential bride price for educated
versus uneducated women, whereby it turns out that they are identical ( p −value of the F -test 46%). While this gives the initial impression that higher
bride price payments reduce the reproductive pressure for uneducated women,
the coeffi cient estimates for educated versus uneducated women are statistically
identical. Nonetheless, women who are economically dependent on their husband
and received a low bride price tend to have more children, suggesting that they
are less valued and have to gain respect through reproductive success (Column 2
of Table 6).
For men, the human capital and income dynamics with respect to bride price are
the opposite of those for women. Educated men and those with income generating
activities expect their wives to have fewer children when a high bride price has
been paid (Columns 3 and 4 of Table 6). However, only the educational difference
is statistically significant ( p− value of the F -test 2%).These findings indicate that poor women and men with low levels of education
are the most negatively affected by the tradition of bride price payments. Poor
women have little bargaining power to ask for high bride price payments, which
limits them in their choice of a husband and might result in less affectionate yet
rather affordable marriage arrangements. Furthermore, being in such a marriage
of convenience puts them under high reproductive pressure.
6.2 Heterogeneity with respect to economic conditions and
wealth
Wealth and economic conditions during the period of the wedding lend themselves
as further possible transmission channels. For households that own land, there is
a negative relationship between the bride price payments and fertility decisions
(Column 1 of Table 7). By contrast, this effect does not exist for households with-
out land ( p− value of the F -test 8%). Hence, women who received a high bride
14
price are valued per se in relatively wealthier households, facing less pressure for
reproductive success and more (economic) independence. In the agrarian society
of rural Senegal, where wealth is determined by the land owned, relatively richer
husbands can afford higher bride prices while putting their wives under less pres-
sure for offspring. Given that these men derive part of their reputation from their
wealth, their prestige depends less on having a large family. However, this is not
to suggest that landowners who can afford the market bride price have no children
at all, rather that they merely have fewer children.
Have bride price dynamics changed over time? The coeffi cients of the interaction
terms between the different year of marriage cohorts and the bride price are not
significant (Column 2 of Table 7). Thus, women are equally valued independent
of the marriage cohort to which they belong. These results demonstrate that
the effect of bride price payments on fertility decisions is stable over time, which
reinforces the robustness of our findings.
6.3 Appreciation of the bride by the husband
In this part, we explore the bride’s appreciation by the husband and her reputation
as other possible channels that explain variations in the bride price and the negative
relationship between bride price payments and the number of children. To this end,
we consider the age at marriage, love versus arranged marriages and divorce.
Age at marriageWe first consider the woman’s age at the time of the marriage, splitting the
sample into brides below and above 20 years of age at the point of their marriage
(Column 1 of Table 8). At first glance, the results suggest that higher bride price
payments for older women reduce their pressure for offspring more than for younger
women. However, the F -test of 16% reveals that we cannot conclude the age of
the wife at marriage to be considered a transmission channel. The dynamics and
expectations linked to the bride price are similar for women who get married young
(below the age of 20) and old (above the age of 20).
Love versus arranged marriageBride price dynamics can be further nuanced by analyzing the nature of the mar-
riage, namely love versus arranged marriage. While we cannot control for love,
15
we have information about the age distance between husband and wife. By con-
trolling for the age distance of the couple, we build on anthropological evidence
that arranged marriages between young girls and older men are not uncommon in
rural Senegal (Foley and Drame, 2012). In Column 2 of Table 8, we present the re-
sults controlling for age distance and its interaction with the bride price payment,
finding a direct effect of age distance: the larger the distance, the more pressure
the woman has to give birth. However, there is a threshold at which a higher age
distance reduces the number of children, marked by when the age gap exceeds 23
years.
In addition, we interact the bride price payments with a dummy equal to one if
the age distance is greater than 10 years. The interaction term further reinforces
the finding that higher bride prices reduce the number of children that a women
has to give birth to in arranged marriages. The interaction between bride price
payments and (almost) no age distance proxies for a love marriage, indicating that
individuals of the same age have a higher likelihood of finding each other and get-
ting married due to their mutual affection. Bride price payments are not needed to
regulate fertility decisions in such love marriages, whereas for arranged marriages
in which a considerably older husband has chosen a young girl, the amount of the
bride price payment is decisive for the woman’s pressure for offspring ( p− valueof the F -test 6%). This finding is in line with Gaspart and Platteau (2010, 2007)
as it shows that bride price payments are less important for love marriages. In a
love relationship, the amount of the bride price loses its power over future family
decisions taken by the couple. However, the bride price is a powerful tool in an
arranged marriage (large age distance). In such cases, if the bride and her family
can negotiate a substantial payment, she will face less pressure for reproductive
success.
DivorceWe further explore the situation of divorced women and it appears that being
a divorced woman has no direct effect on fertility decisions (Column 3 of Table
8). However, these women only represent a small share of the sample (8%) and
therefore the dataset exhibits little variation across women concerning their status
prior to the current marriage. When turning to the differential effect of the bride
price that applies for divorced women, we observe no difference in fertility decisions
16
for women who are married for the first time versus women who had a divorce prior
to the current marriage ( p − value of the F -test 22%). Thus, divorce is not a
transmission channel through which the bride price affects fertility decisions.
6.4 Household context
We complete our analysis by identifying transmission channels at the household
level such as household organization, the wife’s rank and the distance to both the
wife’s and husband’s parents.
Polygamous versus monogamous householdsBeing in a polygamous household increases the number of children that a woman
gives birth to (Column 1 of Table 9). This can be explained by the competition
between wives, particularly with respect to inheritance. The higher the number
of children of any individual wife, the higher the share that she and her children
will inherit following the husband’s death. However, once we jointly control for
the rank of the wife within the household, the positive and significant coeffi cient
estimate associate with being in a polygamous union fades away. Again, compe-
tition is at place, with co-wives competing for the husband’s attention, as well as
resources.
The competition hypothesis is reinforced when we split the sample between polyga-
mous and monogamous households. For women living in a polygamous household,
a lower bride price payment increases fertility pressure, while the bride price pay-
ment has no effect on the couple’s decisions concerning fertility in a monogamous
household ( p− value of the F -test 2%).This finding also indirectly supports the results about love versus arranged mar-
riages: whenever a couple is closer in terms of affection, marriage payments have
no significance and thus do not influence fertility decisions. In turn, if a woman
is just an addition, her value and appreciation is defined by the bride price pay-
ment. We further find that once a woman enters a polygamous household, her
rank within the household neither directly nor indirectly influences her number of
children (Column 2 of Table 9).
Distance to parentsWe also explore the distance to the parents of the wife and the husband. While the
17
first variable has no effect on the fertility decisions of the couple, we find a positive
association between the number of children and the distance to her in-laws. Put
differently, women who live near their in-laws have higher reproductive pressure.
However, the magnitude of the coeffi cient is close to 0 and, despite being precisely
estimated, the coeffi cient has no economic significance (Column 3 of Table 9).
We subsequently split the sample between women living more than 20 kilometers
from their parents and those who live within a radius of 20 km from their parents,
finding no differential effects with respect to the impact of bride price payments
on fertility decisions ( p− value of the F -test 19%). However, when we comparewomen living far from their parents in-law, namely those for whose residence is
located at least 20 kilometers away, to those who live closer, bride price dynamics
with respect to distance appear. In response to high bride price payments, women
close to their parents in-law reduce the number of children considerably more
compared to those who live far away from their in-laws ( p− value of the F -test4%). In other words, fertility pressure is stronger for women who receive a low
bride price and live close to their in-laws compared to those who live far from their
parents in-law. When women live close to their in-laws, they are pressured by the
family and judged according to the number of children they give birth to. Living
further away from the in-laws implies a higher level of (economic) independence
and concomitantly less fertility pressure.
7 Conclusion and Implications
Our results suggest that the bride price system has considerable implications for
women’s bargaining power and independence. A woman’s say in a relationship is
determined by a payment between the groom’s and bride’s families. If the payment
is high, the likelihood of reproductive pressure will be lower. If the payment is
low —and cash constraints in rural Senegal are likely to limit the leeway and thus
choices for young men—the women who are chosen might not be the preferred ones
and fertility pressure increases. Moreover, this system brings about inequalities
in the socio-economic conditions of women, depending on the amount of payment
they received upon their marriage.
A marriage represents a social contract between a man and a woman, including
18
their respective families. This contract can be viewed as insurance in the context
of rural Senegal, whereby both parties have an incentive to keep to it. The roles
of the partners are clearly defined. The husband takes the woman and gives her
economic security. His appreciation of the bride and the value she represents for
him is expressed by the bride price he pays. Whenever the groom is willing to
give a higher bride price, he demonstrates his willingness to treat his wife well and
avoid the risk of divorce, which would correspond with a financial failure of the
investment. In turn, from the bride price payment received, the women can judge
whether she needs to gain reputation within the groom’s family by giving birth
to many children. This simple contract is socially accepted and so strong that
it needs no enforcement mechanisms beyond the local community in which it is
embedded. In fact, the more the husband appreciates his wife, as demonstrated by
the bride price payment, the less he expects her to prove herself through reproduc-
tive success. Furthermore our findings also imply that a low bride price increases
the pressure for women to gain reputation through a large number of children.
While this type of marriage contract seems odd to Western observers, this mu-
tual insurance is prominent in poor rural societies, covering not only the spouses
but also their families. Therefore, efforts to eradicate the tradition of bride price
payments and formalize marriages according to national legal standards might es-
pecially make the women concerned worse off. Women in these poor societies tend
to have lower levels of education than men, as well as less access to information,
which prevents them from seeking (legal) support. Moreover, as the traditional
marriage system reinforces socio-economic differences, bride price payments cannot
be viewed as a means of empowering women; rather, they keep them dependent.
However, bride price payments already have less power over women who are eco-
nomically independent at present. Consequently, a direct change of the bride price
system might be too challenging a task, although economic empowerment will give
leeway to girls, even in traditional societies. The established marriage contract can
only be broken indirectly by giving women more possibilities for economic inde-
pendence.
Bride price payments are not only an African phenomenon, having also gained
increasing importance in China in recent years, while marriage migration for high
bride prices is common (Davin 2005). However, even in China, where bride price
19
payments have exploded in value, adverse effects have been documented for the
migrating women and for men from poor areas. Thus, bride prices are an im-
portant marriage market institution that remains in existence even as countries
and regions develop. However, the identified transmission channels that foster the
continuance of this tradition can be used to counteract. One possible entry point
is family planning programs. On the one hand, our findings suggest that these
programs might be ineffi cient if they do not take into account norms and values
that define women’s behavior with regard to fertility decisions in traditional soci-
eties. At the same time, these programs can be used to educate women. Moreover,
family planning programs that are integrated in broader support structures can
offer women opportunities to increase their (economic) independence such that
the social contract that stipulates the marriage arrangements represents a tool of
empowerment for young couples.
References
Anderson, Siwan. 2003. Why Dowry Payments Declined with Modernization in
Europe but Are Rising in India. Journal of Political Economy, 111(2), 269—310.
Anderson, Siwan. 2007. The Economics of Dowry and Brideprice. Journal of
Economic Perspectives, 21(4), 151—174.
Becker, G.S. 1981. A Treatise on the Family. Harvard University Press.
Beegle, Kathleen, Frankenberg, Elizabeth, & Thomas, Duncan. 2001. Bargaining
Power Within Couples and Use of Prenatal and Delivery Care in Indonesia.
Studies in Family Planning, 32(2), 130—146.
Bergstrom, Ted. 1994. On the Economics of Polygyny.
Bishai, David, & Grossbard, Shoshana. 2007. Far Above Rubies: The Associa-
tion Between Bride Price and Extramarital Sexual Relations in Uganda. IZA
Discussion Papers 2982. Institute for the Study of Labor (IZA).
20
Boserup, E. 1970. Woman’s Role in Economic Development. George Allen and
Unwin Limited.
Botticini, Maristella, & Siow, Aloysius. 2003. Why Dowries? American Economic
Review, 93(4), 1385—1398.
Brown, Philip H. 2009. Dowry and Intrahousehold Bargaining: Evidence from
China. Journal of Human Resources, 44(1).
Cameron, A. Colin, Gelbach, Jonah B., & Miller, Douglas L. 2011. Robust Infer-
ence With Multiway Clustering. Journal of Business & Economic Statistics,
29(2), 238—249.
Cherlin, Andrew, & Chamratrithirong, Aphichat. 1988. Variations in Marriage
Patterns in Central Thailand. Demography, 25(3), 337—353.
Davin, Delia. 2005. Marriage Migration in China: The Enlargement of Marriage
Markets in the Era of Market Reforms. Indian Journal of Gender Studies,
12(2-3), 173—188.
Desai, Sonalde. 1992. Children at Risk: The Role of Family Structure in Latin
America and West Africa. The Population and Development Review, 689—717.
Foley, Ellen E., & Drame, Fatou Maria. 2013. Mbaraan and the Shifting Political
Economy of Sex in Urban Senegal. Culture, Health & Sexuality, 15(2), 121—134. PMID: 23181265.
Gaspart, Frederic, & Platteau, Jean-Philippe. 2010. Strategic Behavior and Mar-
riage Payments: Theory and Evidence from Senegal. Economic Development
and Cultural Change, 59(1), 149—185.
Goldschmidt, Walter. 1974. The Economics of Bride Price Among the Sebei and
in East Africa. Ethnology, 13(3), 313—331.
Kanji, S.S.M., & Camara, F.K. 2000. L’union matrimoniale dans la tradition des
peuples noirs. Logiques juridiques. Harmattan.
21
Papps, I. 1980. For Love or Money?: A Preliminary Analysis of the Economics of
Marriage and the Family. Hobart paper. Institute of Economic Affairs.
Peters, Christina. 2011. Effects of Family Planning and Health Services on
Women’s Welfare: Evidence on Dowries and Intra-Household Bargaining in
Bangladesh. Review of Economics of the Household, 9(3), 327—348.
Platteau, Jean-Philippe, & Gaspart, Frederic. 2007. The Perverse Effects of High
Brideprices. World Development, 35(7), 1221—1236.
Rao, Vijayendra. 1993. The Rising Price of Husbands: A Hedonic Analysis of
Dowry Increases in Rural India. Journal of Political Economy, 101(4), 666—77.
Sharma, Amarendra, & Frijters, Paul. 2009. Groom Price —Female Human Capi-
tal: Some Empirical Evidence. The Journal of Socio-Economics, 38(2), 270—279.
Srinivasan, Sharada, & Bedi, Arjun S. 2007. Domestic Violence and Dowry: Evi-
dence from a South Indian Village. World Development, 35(5), 857—880.
Tertilt, Michele. 2005. Polygyny, Fertility, and Savings. Journal of Political Econ-
omy, 113(6), 1341—1370.
Zhang, Junsen, & Chan, William. 1999. Dowry and Wife’s Welfare: A Theoretical
and Empirical Analysis. Journal of Political Economy, 107(4), 786—808.
22
Table 1: Descriptive statistics
Mean Std. Dev.
Number of children 5.20 3.10Bride price 240,917.5 340,149.2Characteristics of the wifeAge 37.00 12.18Age at marriage 17.85 4.63Literacy 0.16 0.37Income generating activity 0.47 0.5Divorce 0.08 0.27Rank of the wife in polygamous householdFirst wife 0.26 0.44Second wife 0.13 0.34Third or fourth wife 0.02 0.15
Characteristics of the husbandAge 51.03 14.33Age distance 14.89 10.61Literacy 0.44 0.50Income generating activity 0.91 0.29Household characteristicsPolygamous 0.42 0.49Distance to wife’s parents 20.00 62.12Distance to husband’s parents 2.55 29.17EthnicityPular 0.24 0.43Serere 0.18 0.38Manding 0.06 0.23Soninké 0.04 0.21Diola 0.02 0.15Other ethnicity 0.03 0.17ReligionChristian 0.04 0.19Other religion 0.01 0.10Caste 0.18 0.39Landownership 2.51 1.73Year of marriage cohortYear of marriage cohort [0, 5[ 0.13 0.34Year of marriage cohort [5, 10[ 0.16 0.37Year of marriage cohort [10, 15[ 0.15 0.35Year of marriage cohort [15, 20[ 0.14 0.35Observations 2,241
Notes:The reference group for the variable rank of the wife in polygamous
household is monogamously married women. The reference group for the vari-
able ethnicity is Wolof. The reference group for the variable religion is Muslim.
The reference group for the year of marriage cohorts is the cohort [25, .].
23
Table 2: Bride price and fertility: Main results
(1) (2) (3) (4)
Bride price (log) -0.048* -0.041 -0.048** -0.040**(0.028) (0.028) (0.015) (0.013)
Characteristics of the wifeAge 0.500*** 0.511*** 0.498*** 0.508***
(0.026) (0.026) (0.044) (0.042)Age square -0.005*** -0.005*** -0.005*** -0.005***
(0.000) (0.000) (0.001) (0.000)Age at marriage -0.209*** -0.259*** -0.229** -0.281***
(0.060) (0.064) (0.077) (0.063)Age at marriage square 0.003** 0.004*** 0.003* 0.004**
(0.001) (0.001) (0.002) (0.001)Literacy -0.274* -0.325** -0.228 -0.277
(0.147) (0.145) (0.159) (0.166)Characteristics of the husbandAge 0.052* 0.047 0.052* 0.043
(0.029) (0.029) (0.027) (0.026)Age square -0.000* -0.000 -0.000* -0.000
(0.000) (0.000) (0.000) (0.000)Literacy 0.094 0.183* 0.048 0.104
(0.110) (0.111) (0.226) (0.226)Household characteristicsPolygamous -0.081 -0.119
(0.120) (0.142)Observations 2,241 2,241 2,241 2,241Region dummies No No Yes YesSurvey year dummies Yes Yes Yes YesEthnicity, religion and caste No Yes No Yes
Notes: All regressions include a constant. Robust standard errors are in parenthesis. Stan-
dard errors are clustered at the region level in Columns 3 and 4. ∗∗∗, ∗∗, ∗ indicate signifi-
cance at the 1, 5, 10% level, respectively.
24
Table 3: Robustness checks: Controlling for income, wealth and economic condi-tions at the time of the marriage
(1) (2) (3) (4)
Bride price (log) -0.036** -0.034* -0.039** -0.036**(0.014) (0.014) (0.014) (0.014)
Characteristics of the wifeAge 0.499*** 0.497*** 0.506*** 0.443***
(0.040) (0.039) (0.041) (0.073)Age square -0.005*** -0.005*** -0.005*** -0.004***
(0.000) (0.000) (0.000) (0.001)Age at marriage -0.277*** -0.280*** -0.284*** -0.263**
(0.061) (0.061) (0.063) (0.082)Age at marriage square 0.004** 0.004** 0.004** 0.004**
(0.001) (0.001) (0.001) (0.001)Literacy -0.297 -0.306 -0.286 -0.279
(0.162) (0.163) (0.166) (0.172)Income gen. Activity (IGA) 0.243** 0.241**
(0.087) (0.086)Characteristics of the husbandAge 0.045 0.050* 0.048 0.045
(0.025) (0.024) (0.026) (0.028)Age square -0.000 -0.000* -0.000 -0.000
(0.000) (0.000) (0.000) (0.000)Literacy 0.099 0.104 0.109 0.100
(0.225) (0.214) (0.215) (0.219)Income gen. Activity (IGA) 0.063 0.029
(0.107) (0.111)Household characteristicsPolygamous -0.130 -0.148 -0.139 -0.141
(0.146) (0.152) (0.148) (0.148)Landownership 0.113** 0.114*** 0.114***
(0.031) (0.029) (0.027)Observations 2,241 2,241 2,241 2,241Region and survey year dummies Yes Yes Yes YesYear of marriage cohort dummies No No No Yes
All regressions include a constant and controls for religion, ethnicity and caste. Robust standard
errors are in parenthesis. Standard errors are clustered at the region level. ∗∗∗,∗∗ , ∗ indicatesignificance at 1, 5, 10% level, respectively.
25
Table4:Bridebridebyyearofsurvey
All
2009
2011
Difference
Mean
Std.Dev.
Mean
Std.Dev.
Mean
Std.Dev.
Mean
Std.Error
Brideprice
240,917.5
340,149.2
312,202.4
392,210.1
184,095.4
279,506.2
128,106.9***
(14,210.82)
Meanbrideprice263,343.3
276,150.9
Observations
2,241
994
1,247
Notes:Standarderrorinparenthesis.∗∗∗indicatesignificanceatthe1%
level.
26
Table 5: Robustness checks: dependent variable is the mean bride price over theyears
(1) (2) (3) (4) (5)
Mean bride price (log) -0.084** -0.074** -0.087** -0.074** -0.069**(0.034) (0.034) (0.032) (0.025) (0.023)
Characteristics of the wifeAge 0.501*** 0.513*** 0.500*** 0.509*** 0.443***
(0.026) (0.026) (0.042) (0.041) (0.071)Age square -0.005*** -0.005*** -0.005*** -0.005*** -0.004***
(0.000) (0.000) (0.000) (0.000) (0.001)Age at marriage -0.212*** -0.260*** -0.230** -0.282*** -0.264**
(0.061) (0.064) (0.077) (0.063) (0.082)Age at marriage square 0.003** 0.004*** 0.003* 0.004** 0.004**
(0.001) (0.001) (0.002) (0.001) (0.001)Literacy -0.272* -0.322** -0.224 -0.274 -0.276
(0.147) (0.145) (0.159) (0.166) (0.173)Characteristics of the husbandAge 0.051* 0.046 0.051* 0.042 0.044
(0.029) (0.029) (0.026) (0.026) (0.028)Age square -0.000 -0.000 -0.000* -0.000 -0.000
(0.000) (0.000) (0.000) (0.000) (0.000)Literacy 0.092 0.182 0.047 0.103 0.099
(0.110) (0.111) (0.227) (0.227) (0.220)Household characteristicsPolygamous -0.077 -0.115 -0.136
(0.120) (0.142) (0.147)Landownership 0.113***
(0.026)Observations 2,241 2,241 2,241 2,241 2,241Region dummies No No Yes Yes YesSurvey year dummies Yes Yes Yes Yes YesEthnicity, religion and caste No Yes No Yes YesYear of marriage cohort dummies No No No No Yes
Notes: All regressions include a constant. Robust standard errors are in parenthesis. Standard errors are
clustered at the region level in Columns 3 and 4. ∗∗∗, ∗∗, ∗ indicate significance at the 1, 5, 10% level,
respectively.
27
Table 6: Bride price and fertility: Human capital and income of the wife and thehusband
(1) (2) (3) (4)Characteristics of the wifeLiteracy 0.128 -0.293 -0.273 -0.280
(0.645) (0.170) (0.172) (0.173)Bride price x Literacy -0.066
(0.044)Bride price x No literacy -0.029*
(0.013)Income gen. Activity (IGA) -0.505
(0.318)Bride price x IGA 0.001
(0.021)Bride price x No IGA -0.066**
(0.021)Characteristics of the husbandLiteracy 0.102 0.093 0.969* 0.101
(0.218) (0.219) (0.404) (0.217)Bride price x Literacy -0.083***
(0.014)Bride price x No literacy -0.006
(0.021)Income gen. Activity (IGA) 0.163
(0.621)Bride price x IGA -0.037*
(0.016)Bride price x No IGA -0.028
(0.050)Observations 2,241 2,241 2,241 2,241Cohort, region and survey year dummies Yes Yes Yes YesF-test (p-value) 0.46 0.04 0.02 0.88
All regressions include a constant and controls for ethnicity, religion, caste, polygamy,
landownership, age variables of the wife and husband. Robust standard errors are in paren-
thesis. Standard errors are clustered at the region level. ∗∗∗,∗∗ ,∗ indicate significance at 1,5, 10% level, respectively. Detailed results available upon request.
28
Table 7: Bride price and fertility: Economic conditions and wealth
(1) (2)Landownership 0.163*** 0.115***
(0.043) (0.027)Bride price x Landownership -0.042**
(0.015)Bride price x No Landownership -0.003
(0.025)Cohort [0, 5[ -0.393
-1.002Cohort [5, 10[ 0.055
(0.728)Cohort [10, 15[ -0.379
-1.270Cohort [15, 20[ -0.168
(0.589)Cohort [20, 25[ 0.759
-1.375Bride price x Cohort [0, 5[ -0.027
(0.041)Bride price x Cohort [5, 10[ -0.034
(0.044)Bride price x Cohort [10, 15[ 0.010
(0.053)Bride price x Cohort [15, 20[ 0.008
(0.049)Bride price x Cohort [15, 20[ -0.055
(0.089)Observations 2,241 2,241Region and survey year dummies Yes YesF-test (p-value) 0.08 0.08
All regressions include a constant and controls for ethnicity, reli-
gion, caste, polygamy, age and literacy variables of the wife and
husband. The reference group for the year of marriage cohorts
is the cohort [25, .]. Robust standard errors are in parenthesis.
Standard errors are clustered at the region level. ∗∗∗,∗∗ ,∗ indi-cate significance at 1, 5, 10% level, respectively. Detailed results
available upon request.
29
Table 8: Bride price and fertility: Appreciation of the bride by the husband
(1) (2) (3)Below 20 -0.455
(0.344)Bride price x Below 20 -0.015
(0.019)Bride price x Above 20 -0.081**
(0.029)Age distance 0.046*
(0.020)Age distance square -0.001**
(0.000)Bride price x High age distance -0.039**
(0.015)Bride price x No age distance -0.023
(0.016)Divorce 0.383
(0.977)Bride price x divorce -0.244
(0.177)Bride price x No divorce -0.019
(0.011)Observations 2,241 2,241 2,241Cohort, region and survey year dummies Yes Yes YesF-test (p-value) 0.16 0.06 0.23
All regressions include a constant and controls for ethnicity, religion, caste, polygamy,
landownership, age and literacy variables of the wife and husband. Robust standard
errors are in parenthesis. Standard errors are clustered at the region level. ∗∗∗,∗∗ ,∗
indicate significance at 1, 5, 10% level, respectively. Detailed results available upon
request.
30
Table 9: Bride price and fertility: Household context
(1) (2) (3)Polygamy 1.171*** -0.303 -0.141
(0.304) (0.586) (0.147)Bride price x polygamy -0.291**
(0.102)Bride price x No polygamy -0.015
(0.008)Rank of the wife in polygamous householdFirst wife 0.448
(0.605)Second wife 0.601
-1.042Third or fourth wife 0.580
-1.199Bride price x First wife -0.020
(0.069)Bride price x Second wife -0.041
(0.057)Bride price x Third or fourth wife -0.087
(0.075)Distance to wife’s parents 0.000
(0.001)Bride price x Distance to wife’s parents -0.042
(0.029)Bride price x No distance to wife’s parents -0.003
(0.003)Distance to husband’s parents 0.003*
(0.001)Bride price x Distance to husband’s parents -0.400**
(0.155)Bride price x No distance to husband’s parents -0.036**
(0.013)Observations 2,241 2,241 2,241Cohort, regions and survey year dummies Yes Yes YesF-test (p-value )distance to wife’s parents 0.03 0.88 0.19F-test (p-value) distance to husband’s parents 0.04
All regressions include a constant and controls for ethnicity, religion, caste, polygamy,
landownership, age and literacy variables of the wife and husband. Robust standard er-
rors are in parenthesis. Standard errors are clustered at the region level. ∗∗∗,∗∗ ,∗ indicatesignificance at 1, 5, 10% level, respectively. Detailed results available upon request.
31
